Every row of the evidence table

Sustained mild and moderate drinking carried lower risk than not drinking, sustained heavy drinking higher risk, and cutting down from heavy to moderate lower risk. Observational, with the same abstainer problem.

U-shaped: long-term abstainers had 47% higher risk than moderate drinkers, and above 14 units a week each extra 7 units raised risk by 17%. Neither end of the range looks like a benefit of drinking.

The two measures moved in opposite directions over 2015-2021: standardised incidence fell from 3.5% to 2.8% per year while prevalence rose from 10.5% to 11.8%; gaps between Black or Hispanic and White beneficiaries narrowed.

The review's own summary points two ways: CSF biomarker tests were mostly cost-effective against cognitive assessment or imaging, while blood biomarker tests showed mixed results; every included study was model-based.

The findings point two ways: the p-tau217/Aβ42 ratio tracked AD-related factors (MCI, AD dementia, future progression, age, APOE ε4) more closely than either marker alone and had few intermediate results, but it was also related to kidney function, diabetes and hearing loss, and medical morbidities were most frequent among intermediate classifications.

Direction is for the test's accuracy against the reference standard; an accuracy result, not a health outcome. Performance varied widely across tests, from 49.3% to 91.4% pooled sensitivity and 61.5% to 96.7% pooled specificity at a single cut-point, with most studies at high risk of bias.

The results point different ways: diagnostic confidence rose 1.43 points, but diagnosis changed in only 17.2% and management in 27.7%, below the 30% threshold the investigators set as clinically significant; a non-elevated result reduced amyloid PET use, and patients with elevated results had higher anxiety at 2-3 months.

The findings point different ways: lower eGFR raised every blood biomarker except Aβ42/40 (a test-interpretation caveat), yet impaired kidney function did not itself raise dementia risk (HR 0.93), while the association between high NfL and dementia was about twice as strong in people with impaired kidneys.

The two measures behaved differently: baseline plasma p-tau217 predicted later cognitive decline about as well as entorhinal tau PET (ρ -0.47 vs -0.55), but over time p-tau217 rose then decelerated and tracked concurrent cognitive change only weakly (ρ -0.24) or not at all in the PET substudy, whereas regional tau PET change correlated strongly with concurrent decline.

The three markers pointed different ways: plasma p-tau217 distinguished amyloid/tau stages, GFAP tracked amyloid-driven tau progression, and NfL was only weakly linked to Alzheimer pathology and, in the authors' words, inadequate as a neuronal injury biomarker for AD.

The result points two ways: p-tau181 gave the best AUC for separating AD from controls, frontotemporal dementia and amyloid-negative Lewy body dementia, but that discrimination could not be reproduced when pre-defined cut-offs were applied, so cohort-specific cut-offs were needed.

Direction is for the test's accuracy against the reference standard; an accuracy result, not a health outcome. Results ranged from 55% to 100% sensitivity and 60% to 100% specificity across four laboratories using different protocols, with only fair agreement between them (kappa 0.246).

Peer support reduced distress to behavioural symptoms and increased social support and quality of life immediately after the intervention, but had no effect on depression/anxiety, loneliness, burden or competence, and no outcome was sustained at follow-up.

The two comparisons point different ways: overall utilization and payments were similar between amyloid-positive (A4) and amyloid-negative (LEARN) participants, but among A4 participants claims indicators of progression carried 45-103% higher payments.

The review splits by drug class: at $100,000 per QALY every study found memantine cost-effective and more than half found the cholinesterase inhibitors so, whereas aducanumab, lecanemab and donanemab were not cost-effective in any study. A value result, not a clinical harm.

Direction is for value for money: APPLE-Tree had a 96% probability of being cost-effective per cognition point at a £30 000 threshold but only 12% on QALYs, which were slightly lower in the intervention arm. Post hoc subgroups suggested better value in non-White and socioeconomically disadvantaged participants.

The two perspectives disagree: the digitally supported care management system dominated usual care for the public payer (lower costs, marginally more QALYs, 72-79% probability of cost-effectiveness) but cost €331,000 per QALY from a societal perspective (33-35% probability).

MCI was diagnosed sooner (HR 1.21) but dementia diagnosis barely moved (HR 1.04, 95% CI 1.02-1.06). The outcome is detection, not health.

Admission showed no evidence of an effect on 30-day mortality or subsequent inpatient days, but raised 30-day health care spending by $2547. An instrumental-variable estimate for marginal admissions.

Advance care planning strongly increased DNR documentation (OR 14.12) but did not change whether residents died in the nursing home (OR 1.62, 95% CI 0.72-3.66). Evidence on other end-of-life outcomes was limited and inconsistent, and no study came from a non-Western setting.

In dementia there may be small-to-moderate benefits at six months (low-certainty evidence, I2 88-96%); in MCI ginkgo probably has little or no effect (moderate certainty); in subjective cognitive complaints the effect is very uncertain. Adverse events were similar to placebo.

The biochemical primary outcome was met (DHA raised the CSF DHA-to-arachidonic-acid ratio versus placebo), but there was no difference in cognition or brain volume over 24 months. The authors recommend against further supplementation trials of this kind.

The overall PM2.5 estimate includes no effect (HR 1.04, 95% CI 0.99-1.09). Only the studies that actively assessed everyone for dementia showed a clear effect (HR 1.42, 95% CI 1.00-2.02). Most studies were at high risk of bias.

High overall and leisure-time activity were associated with lower dementia risk, but occupational and commuting activity were associated with higher risk. Certainty of evidence was very low to low.

Direction is for APOE genotype after 90: ε2 carriers had lower dementia risk, while the ε4 association was not statistically significant. Women and Black participants had higher risk than men and Asian participants. Age-standardised incidence was 116.82 per 1000 person-years.

The intervention-control cognitive difference was statistically significant in the higher-genetic-risk halves and not in the lower halves, but neither genetic score significantly modified the intervention effect overall (P > 0.46); an exploratory sex difference emerged for the Alzheimer score.

The study's own results point different ways: the polygenic score added to APOE for predicting AD/MCI and for progression from amyloid-positive MCI to AD, but amyloid deposition was predicted by APOE alone.

The two kinds of variant point opposite ways: deletions and loss-of-function alleles in ABCA1 and ABCA7 and 22q11.21 deletions raised AD risk, while 22q11.21 duplication lowered it (odds ratio 0.34 at exome-wide significance in replication). Not modifiable.

A statistically detectable but very small cognitive effect (ADAS-Cog SMD -0.11) alongside 107 more cases of ARIA-E per 1,000. The reviewers' own conclusion is that removing amyloid does not seem to produce clinically meaningful effects — the sharpest disagreement in this table with how the individual trials were reported.

CHMP refused the drug, then on re-examination recommended it only for people with one or no APOE4 copy, where the ARIA rates are lower. Benefit on cognition and harm from ARIA in the same document, with the restriction as the regulator's way of balancing them.

Benefits were statistically present but every one fell below the minimum clinically important difference, against a number needed to harm of 9 for ARIA-oedema. Agrees with the Cochrane review and not with the trial reports.

Approved on amyloid reduction (surrogate) while reporting higher ARIA rates than placebo; the notice reports no cognitive outcome.

Two identically designed trials disagreed: EMERGE met its primary endpoint, ENGAGE did not.

Antihypertensive treatment improved cognitive scores by a small amount (SMD 0.06, moderate certainty) but the reduction in dementia incidence was not statistically significant (OR 0.89, low certainty). The drug-class rankings rest on one trial per class and are described as highly uncertain.

Antihypertensive use tracked lower Alzheimer's risk but not lower non-Alzheimer dementia risk, where treated and untreated hypertension looked the same. A single baseline blood pressure reading predicted neither.

The split inside this one paper is the finding: the 41 observational studies show large benefits (SGLT2i OR 0.56, GLP-1RA OR 0.58) while the 23 randomised trials show no difference. When design and result disagree this cleanly, the trials are the ones to believe.

Two findings point opposite ways: synchronised norepinephrine, blood-volume and CSF oscillations during natural NREM sleep were the strongest predictors of glymphatic clearance, while the sleep aid zolpidem suppressed those oscillations and glymphatic flow. Mouse work with no human outcome.